基于集合预报的中国极端强降水预报方法研究

极端强降水天气属于小概率事件,其发生具有很多不确定的因素,预报难度很大。根据Anderson-Darling检验原理研究基于集合预报资料的极端强降水天气预报方法,利用2007—2010年中国T213集合预报资料和2001—2010年6—8月中国降水观测资料,分析观测与集合预报累积概率密度分布函数的特征,建立基于集合预报与模式历史预报累积概率密度分布函数连续差异的数学模型——极端降水天气预报指数(EPFI),并对2011年7月中国极端强降水天气进行预报试验。结果表明,极端降水天气预报指数可以充分利用集合降水累积概率密度分布的尾端信息,为极端强降水提供科学合理的预报,基于中国气象局(CMA) T213集合预报的极端降水天气预报指数可提前3—7 d发出极端强降水预警信号,随着预报时效的延长,预报技巧逐渐降低。研究还表明,模式气候累积概率分布的合理性将直接影响极端强降水天气识别能力。     

冬季稳定性降水相态预报研究进展

冬季降水无论对地面的生产生活还是对高空飞机航行都可能造成严重灾害,降水相态预报的准确性决定了冬季降水预报的成功,该文系统回顾了近几十年降水相态预报取得的成果。降水相态预报方法大致分为3类:第1类是基于观测或数值天气预报建立的指标以及回归方程,其中某些方法高度依赖数值天气预报模式准确率;第2类是基于数值天气预报模式的微物理方案法和集合预报法;第3类是基于观测和数值预报产品的人工智能预报法。近年来降水相态模式预报产品准确率不断提高,成为降水相态预报中一个重要的产品支撑。但如何将降水相态形成机制的微物理研究成果用于改善数值预报模式降水相态预报的技巧,以及如何利用人工智能等技术提高降水相态预报的准确率等方面还需要不断努力。     

降雨型滑坡的集合预报模型及其初步应用的试验研究

滑坡的实时预警系统GRAPES-Landslide是将数值天气预报模式GRAPES (Global/Regional Assimilation and PrEdiction System)与滑坡预测模型TRIGRS (Transient Rainfall Infiltration and Grid-based Regional Slopestability)进行单向耦合建立起来的动力数值预报预警系统。由于滑坡预测模型TRIGRS中的关键水土参数具有空间分布很不均的特性,很难获取准确的数据加以描述,使得滑坡事件的激发、预测存在很大的不确定性,同时数值天气预报模式本身具有不确定性,因而用于诱发滑坡灾害的估测降水存在不确定性,进而使得滑坡的预报存在偏差。本研究基于预测降水和水土参数分布不确定性的考虑,提出了GRAPES-Landslide滑坡集合预报模型。滑坡集合预报模型中有5个不同的预报降水成员,分别是(1) GRAPES_MESO业务模式、(2)"暖-潜热加热纳近方法、(3)基于九点平滑滤波的"暖-潜热加热纳近"方法、(4)对(1)~(3)的降水成员进行简单平均、(5)对(1)~(3)的降水成员进行概率匹配的集合。根据水土参数呈正态分布的特点,通过Monte-Carlo方法随机生成100组扰动参数值。将5个预报降水与100组扰动水土参数结合,组成GRAPES-Landslide滑坡集合预报模型。选择2013年7月18日00时到7月19日12时(协调世界时)福建省"西马仑"台风降雨引发闽三角地区发生大量滑坡灾害为例,进行实际预报试验。初步研究结果表明本文建立的GRAPES-Landslide滑坡集合预报系统所预测的滑坡频发区与观测区域有很好的吻合度,与目前的滑坡业务预报结果相比有明显改进,落区更精细化。因此,GRAPES-Landslide滑坡集合预报系统综合考虑了降水预报的不确定性和非均匀分布的水土参数的不确定性,为区域滑坡预测提供了一种新的可能方法。     

基于交叉验证的多模式超级集合预报方法研究

利用AREM、MM5和WRF 3个中尺度有限区域模式,通过选取对短期天气预报影响颇大的积云参数化方案和边界层方案构成15个集合预报成员,以2003年7月汛期天气为研究对象,分别采用相关加权、多元线性回归以及支持向量机回归与"交叉验证"相结合的方法,开展有限区域模式的多模式短期超级集合预报研究.文中主要对上述3种方法的24 h降水和700 hPa流场的超级集合预报结果与多模式集合平均预报结果以及T213模式结果进行了对比分析,结果表明:(1)对于24 h降水,支持向量机回归方法的超级集合预报得到的均方根误差比多模式集合平均小,各降水临界值的TS异常评分比多模式集合平均高;并且它也较相关加权法和多元线性回归的超级集合预报效果好.(2)对于700 hPa流场,对比分析各预报结果经过向量EOF分析得到的风场第1模态和第2模态表明,多模式集合平均主要使风场强度变小,多元线性回归和支持向量机回归的超级集合预报可以较好地刻画风场的强度分布,其中支持向量机回归的超级集合预报对风场强度及其区域分布的预报效果最好.(3)对于700 hPa流场,超级集合预报明显优于同期T213模式预报,从相同的预报均方根误差意义看,支持向量机回归的超级集合预报至少较T213模式预报能提前12 h.     

极端降水天气预报指数对气候累积概率分布敏感性研究

针对中国气象局T213全球集合预报系统历史样本较短、极端降水天气预报指数数学模型中气候累积概率分布样本不足的问题,选取气候背景和地理地貌特征较为相似的长江中下游地区,利用扩展时间序列和空间范围的方法,增加T213全球集合预报系统的降水预报气候分布样本数,改进模式降水预报气候累积概率分布,利用2011年6月10—20日集合预报资料进行极端降水天气预报指数预报试验,分析气候累积概率分布敏感性。结果显示:扩展时间序列和空间范围增加模式T213集合预报气候样本数的方法,生成的模式气候累积概率分布较单一格点方法更具代表性,能提高极端降水天气预报指数的识别技巧,并提前8 d发出长江中下游地区极端强降水预报信号。     

青藏高原东侧"2003.8.28"暴雨的集合预报试验

利用MM5模式和国家气象中心的T213模式的预报资料,通过研究非绝热物理过程参数化方案对高原东侧"2003.8.28"暴雨数值预报的影响特征,进行了多物理模式集合预报试验,为开展青藏高原东侧集合预报扰动技术研究进行了试验.试验结果表明,模式物理参数化方案对中尺度降水预报结果有明显影响,包括局地降水强度、空间分布型态、时间演变特征等.随着模式分辨率的提高,积云对流参数化方案将增加小雨量级降水区域,产生一些虚假降水,就现阶段模式水平而言,高分辨率集合预报应重点发展考虑强降水预报不确定性的集合预报模式系统.多物理模式集合预报的初步试验结果表明,高分辨率集合预报可以改进单一确定性预报结果不稳定的缺点,为强降水灾害性天气预报提供有价值的预报信息.     

上海区域数值预报模式集合预报系统的建立与试验

以目前运行的上海区域业务数值预报模式为基础,从预报模式的不确定性出发构造8个预报成员,建立了上海区域数值预报模式集合预报系统的初步模型,并对2004年夏季进行了逐日48 h预报试验。结果表明:集合平均对华东地区城市降水、温度、海平面气压等气象要素的总体预报能力与分辨率高3倍的业务模式相当,其中对雨量较大降水、最低温度、海平面气压(0~24 h)的预报效果好于业务模式;集合预报还能提供客观化、定量化的降水概率预报,对降水的发生、尤其是特大降水的发生有着很好的提示作用。     

地面气象要素多模式集成预报研究进展

目前,集合预报已成为天气预报业务的主要支撑。然而,由于数值模式本身的限制与不完善以及集合系统存在初值扰动、集合大小等方面的局限,常存在预报偏差。不同预报模式通常具有不同的物理过程参数化方案、初始条件等,导致其预报能力各有不同。为此,如何纠正预报偏差以及如何充分有效地利用不同模式的预报信息以获得更加准确的天气预报广受关注。近年来,利用统计理论与预报诊断,基于多个集合预报系统的多模式集成预报技术得到快速发展,已成为有效消除预报偏差从而提高天气预报技巧的一种统计后处理方法。针对气温、降水和风3个最基本的地面气象要素,首先依据预报形式将应用范围较广的简单集合平均、消除偏差集合平均、超级集合、贝叶斯模式平均、集合模式输出统计等加权或等权平均多模式集成技术,分成确定性预报和概率预报两大类,并做系统介绍。最后,讨论使用和发展多模式集成技术需要关注的问题,包括考虑参与集成的模式个数、发展降水及风速分级预报模型和发展基于机器学习的多模式集成新技术。     

2003年江淮汛期多模式短期集合预报方法研究

利用AREM、MM5和WRF模式为试验模式,由对短期天气预报结果影响颇大的积云参数化方案和边界层方案构成15个集合预报成员,开展有限区域多模式短期集合预报在我国汛期时段的应用与研究.分别研究了单个模式集合预报和多模式集合预报在2003年汛期(7月)预报中的应用,预报对象主要包括降水、500 hPa位势高度和700 hPa相对湿度.试验结果表明:(1) 由AREM、MM5和WRF模式构成的多模式集合对以上要素的集合预报总体效果比其任一单个模式的集合预报效果好;(2) 对于降水的集合预报,单个模式的集合平均结果对多模式集合预报效果有影响.且对于不同的降水临界值影响不同;当降水临界值较小时,单模式集合平均结果对多模式集合效果影响较小;当降水临界值较大时,影响较大,甚至可以影响多模式集合的集合平均预报成败;(3) 对于降水、500 hPa位势高度和700 hPa相对湿度,其单个模式以及多模式的48 h集合预报对确定性预报的改善度都比24 h的显著.(4) 对于形势预报和相对湿度预报,多模式集合预报效果明显比同期T213模式的预报水平高.     

海河流域东北冷涡背景下的降水预报订正研究

针对海河流域东北冷涡降水样本,应用海河流域加密自动站降水资料及欧洲中期天气预报中心（ECMCWF）降水预报资料,利用滑动相关分析方法建立重组预报序列,基于加密自动站24 h累积降水量及重组24 h降水预报序列的Gamma累积概率分布曲线,采用预报—实况概率匹配方法建立1～3日的短期订正模型并进行试报检验。结果表明:欧洲中心数值模式对于海河流域东北冷涡降水的预报较实况偏慢;概率匹配法主要通过订正降水量级来改善预报结果,订正后降水预报对于小雨、大雨、暴雨预报的TS（Threat Score）评分技巧均有提升,尤其对于大雨和暴雨及以上量级预报,订正后预报量级及预报落区大小均与实况更加接近,订正效果显著。东北冷涡降水对流性强,模式预报能力弱,而订正后预报能有效提高此类强降水的预报技能,具有较好的应用价值。     

ConvLSTM Based Temperature Forecast Modification Model for North China

The correction of model forecast is an important step in evaluating weather forecast results. In recent years, post-processing models based on deep learning have become prominent. In this paper, a deep learning model named ED-ConvLSTM based on encoder-decoder structure and ConvLSTM is developed, which appears to be able to effectively correct numerical weather forecasts. Compared with traditional post-processing methods and convolutional neural networks, ED-ConvLSTM has strong collaborative extraction ability to effectively extract the temporal and spatial features of numerical weather forecasts and fit the complex nonlinear relationship between forecast field and observation field. In this paper, the post-processing method of ED-ConvLSTM for 2 m temperature prediction is tested using The International Grand Global Ensemble dataset and ERA5-Land data from the European Centre for Medium-Range Weather Forecasts (ECMWF). Root mean square error and temperature prediction accuracy are used as evaluation indexes to compare ED-ConvLSTM with the method of model output statistics, convolutional neural network postprocessing methods, and the original prediction by the ECMWF. The results show that the correction effect of ED-ConvLSTM is better than that of the other two postprocessing methods in terms of the two indexes, especially in the long forecast time.     

Current Status and Future Challenges of Weather Radar Polarimetry: Bridging the Gap between Radar Meteorology/Hydrology/Engineering and Numerical Weather Prediction

After decades of research and development, the WSR-88 D(NEXRAD) network in the United States was upgraded with dual-polarization capability, providing polarimetric radar data(PRD) that have the potential to improve weather observations,quantification, forecasting, and warnings. The weather radar networks in China and other countries are also being upgraded with dual-polarization capability. Now, with radar polarimetry technology having matured, and PRD available both nationally and globally, it is important to understand the current status and future challenges and opportunities. The potential impact of PRD has been limited by their oftentimes subjective and empirical use. More importantly, the community has not begun to regularly derive from PRD the state parameters, such as water mixing ratios and number concentrations, used in numerical weather prediction(NWP) models.In this review, we summarize the current status of weather radar polarimetry, discuss the issues and limitations of PRD usage, and explore potential approaches to more efficiently use PRD for quantitative precipitation estimation and forecasting based on statistical retrieval with physical constraints where prior information is used and observation error is included. This approach aligns the observation-based retrievals favored by the radar meteorology community with the model-based analysis of the NWP community. We also examine the challenges and opportunities of polarimetric phased array radar research and development for future weather observation.     

A Deep Learning Method for Bias Correction of ECMWF 24–240 h Forecasts

Correcting the forecast bias of numerical weather prediction models is important for severe weather warnings. The refined grid forecast requires direct correction on gridded forecast products, as opposed to correcting forecast data only at individual weather stations. In this study, a deep learning method called CU-net is proposed to correct the gridded forecasts of four weather variables from the European Centre for Medium-Range Weather Forecast Integrated Forecasting System global model(ECMWF-IFS): 2-m temperature, 2-m relative humidity, 10-m wind speed, and 10-m wind direction, with a forecast lead time of 24 h to 240 h in North China. First, the forecast correction problem is transformed into an image-toimage translation problem in deep learning under the CU-net architecture, which is based on convolutional neural networks.Second, the ECMWF-IFS forecasts and ECMWF reanalysis data(ERA5) from 2005 to 2018 are used as training,validation, and testing datasets. The predictors and labels(ground truth) of the model are created using the ECMWF-IFS and ERA5, respectively. Finally, the correction performance of CU-net is compared with a conventional method, anomaly numerical correction with observations(ANO). Results show that forecasts from CU-net have lower root mean square error, bias, mean absolute error, and higher correlation coefficient than those from ANO for all forecast lead times from 24 h to 240 h. CU-net improves upon the ECMWF-IFS forecast for all four weather variables in terms of the above evaluation metrics, whereas ANO improves upon ECMWF-IFS performance only for 2-m temperature and relative humidity. For the correction of the 10-m wind direction forecast, which is often difficult to achieve, CU-net also improves the correction performance.     

A case study of excessive rainfall forecasting

Summary Flash floods have been recognized as one of the most significant natural disaster problems in the world. Within the United States, the annual average flood death toll exceeds one hundred and property damage is on the order of a billion dollars. There has been an increased effort of the meteorological community to improve short term quantitative precipitation forecasting, principally by improving mesoscale numerical weather prediction for heavy rain events. Nevertheless, to date, numerical weather prediction has had rather limited impact on the prediction of the most damaging convective rainstorms.This study examines numerical experiments, including both coarse-mesh and fine-mesh model simulations, of the Enid, Oklahoma flood of 10–11 October 1973. Besides the great concentration of rainfall, the Enid flood was rather unique in comparison with other flash flood cases in that it was part of a much larger area of heavy rainfall which soaked the central Plains over the 24h period ending at 1200 UTC 11 October. The objective is to assess the overall usefulness and limitation of numerical weather prediction models in quantitative precipitation forecasting for this flash flood event.The model experiments reveal that the broad-scale precipitation patterns associated with the front and cyclone are well predicted, but the maximum rainfall amounts around Enid are underpredicted. The fine-mesh model is superior to the coarse-mesh model because of the former's ability to generate many significant mesoscale features in the vicinity of the front. In the fine-mesh model, many convection-related parameters (e.g., moisture flux convergence) are correlated very well temporally and spatially with the observed heavy precipitation scenario.     

BP神经网络在油菜花期预报中的应用

建立花期预报模型,发布观赏性植物的精准花期预报,为旅游活动提供重要参考依据,已经成为气象服务领域一个新的发展方向。为了解高淳旅游区油菜花期的变化规律,探索其预报方法,指导高淳油菜花节旅游活动,根据1985—2010年高淳站日最高气温、日最低气温、日平均气温、日降水量、日日照时数、日平均5 cm地温、日平均相对湿度和日小型蒸发量等气象观测数据,利用主成分分析法,得到其与油菜花期相关系数较大的3个主成分,即温度因子、天气因子和辐射因子,以此为输入因子,建立基于BP神经网络的油菜花期预报模型,探讨BP神经网络在花期预报领域的应用。结果表明,传统的有效积温方法预报结果与实际开花期平均相差4.25天,BP神经网络方法预报结果与实际开花期平均相差1.5天,与有效积温预报油菜花期的方法相比,BP神经网络技术具有预测结果准确率高和操作简单等特点,在花期预报领域具有广阔的应用前景。     

Numerical weather prediction over the Eastern Caribbean using Florida State University (FSU) global and regional spectral models and multi-model/multi-analysis super-ensemble

Summary A month-long short-range numerical weather prediction experiment using the Florida State University’s (FSU) global and regional models and the multi-model/multi-analysis super-ensemble over the Eastern Caribbean domain is presented in this paper. The paper also investigates weather prediction capabilities of FSU global and regional models by examining the root mean square errors (RMSE) for the wind and precipitation fields. Super-ensemble forecasting, a new statistical approach to weather forecasting, is used over this domain. Here, forecasts from a number of numerical models provide the input and statistical combinations of these forecasts produce the super-ensemble forecast. A similar approach is used for the precipitation field where one model using different rain rate algorithms is used to generate different model outputs. The results show that the super-ensemble method produces forecasts that are superior to those obtained from the ensemble members. Received May 29, 2000/Revised February 15, 2001     

基于统计降尺度方法的夏季降水精细化预报

利用1999—2009年安徽省淮河以南地区60个县市站夏季逐日降水资料和安庆市探空站逐日资料,研究了中低层不同风向配置下局地降水与大尺度降水场之间的关系,以3种不同预报对象及相应的预报因子分别采用神经网络和线性回归方法设计6种预报模型对观测资料进行逼近和优化,从而实现空间降尺度.分析对比6种预报模型46站逐日降水量的拟合和预报效果,结果表明:采取相同的预报对象及预报因子的BP神经网络模型在拟合和预报效果上均好于线性回归模型,可见夏季降水场之间以非线性相关为主;神经网络模型预报结果同常用的Cressman插值预报相比,能很好地反映出降水的基本分布及局地特征;预报对象为单站降水序列的神经网络模型在以平原、河流为主要地形的区域预报效果较好,预报对象为REOF主成分的神经网络模型则在山地和丘陵地形区域预报效果较好.